A mini tool to better handle your Conda environment and Python package

About

Conda-piptools makes your Conda environment and Python package management easier:

• It allows you to specify easily the exact Python version as well as other things you want (e.g. CUDA and CUDNN versions) for a Conda environment;
• It allows you to better handle the dependencies and the compatibilities of your Python packages. More precisely, it requires you to only specify minimal constraints (e.g. torch >= 1.7 and numpy) and Conda-piptools will then figure out exact and mutually compatible versions (e.g. torch==1.7.1 and numpy==1.19.5). `

Note

As its name implies, Conda-piptools uses Conda and pip-tools in the background.

Requirements

You need to have Conda installed.

Usage

Follow these steps:

1. Specify your Python version as well as the versions of other things (e.g. CUDA and CUDNN versions) you want for a conda environment in the environment.yml file;
2. Use the setEnvName.sh script to set the name of the new or existing conda environment:

./setEnvName.sh "yourEnvironmentName"

3. Specify your Python package requirements for the development and the production respectively in requirements/dev.in and requirements/prod.in;
4. Use the make command to create and/or update all your Python packages (use make force if you want to force it):

make

Multiple Conda installed

If you have multiple Conda installed, use Conda-piptools inside the base Conda environment of the targeted Conda.